Skip to main content

Work Process and Memory Management in SAP

Let’s talk about the entire concepts that are related to memory when we talk about SAP Application.



Starting with few basic terminologies,


Local Memory : 


Local process memory, the operating system keeps the two allocation steps transparent. The operating system does the other tasks, such as reserving physical memory, loading and unloading virtual memory into and out of the main memory.



Shared Memory : 


If several processes are to access the same memory area, the two allocation steps are not transparent.


One object is created that represents the physical memory and can be used by various processes. The processes can map the object fully or partially into the address space. The way this is done varies from platform to platform. Memory mapped files, unnamed mapped files, and shared memory are used. 


Extended Memory :


SAP extended memory is the core of the SAP memory management system. Each SAP work process has a part reserved in its virtual address space for extended memory. You can set the size of extended memory using the profile parameter em/initial_size_MB: Extended Memory Pool Size.


Mapping can be done from any extended memory from the common resource to any work process.


[If you have programming understanding , It is majorly used when pointers are involved in ABAP Program]


SAP Swap Memory :


Allocation of memory for the current internal session by transferring pages out of memory, similarly to operating system paging. SAP Paging enables the roll area to be extended at ABAP runtime when a large dataset, internal tables is handled.




Virtual Memory :


Virtual memory is fully independent of the physical main memory.


All operating systems which we use for SAP Applications support virtual memory technology. 


A process allocates virtual memory using logical (virtual) addresses. Each process has its own virtual address space 


Lets understand what is Address Space, The address space consists of a text segment, a data segment, a dynamically extendible heap, and a dynamically extendible stack. 


About Text segment (a.k.a code segment) and data segment. 


Text segment is where you store the program.

Data segment is where you save the data.


About Heap and Stack Memory Allocation :


The heap increases in size from the bottom , The memory is allocated during the execution of instructions written by programmers.

The stack increases from the top and  allocation happens on contiguous blocks of memory, which enables the entire virtual address space to be used. 


This also applies to SAP work processes


Understand it this way , 


With 32-bit systems, 4 GB of memory is theoretically available. Depending on the operating system, there is approximately 2 GB of virtual address space available to a process.


We reserve heap area for SAP Work Processes (we have profile parameters to set them up)


  1. SAP Roll Area : The roll area is area in memory which belongs to a work process(we can configure that ). It is located in the heap of the virtual address space of the work process.


When the context of a work process changes (when the previous work process is completed and new one task take up to that work process ), the data is copied from the roll area to a common resource known as the roll file. To prevent repeated copying, another roll buffer is located in between, which is part of the shared memory.


We have two segments in roll area, 


ztta/roll_first : assigned to the work process first as memory


ztta/roll_area : if the memory that was assigned earlier is used and the work process needs more memory then this memory is assigned.


The amount of memory is determined by the difference between the ztta/roll_area and ztta/roll_first parameters.


Roll area holds data from previous dialog steps and data that characterize the user.


 

  1. SAP Paging Area :  The paging area stores special data for the ABAP processor, while the extended memory stores a large portion of the internal tables of a program.


Parameter which determines this is rdisp/PG_MAXFS


Paging area holds data from the application programs.


What is stored in Paging Area here is an example :


    The ABAP/4 runtime system and the ABAP/4 compiler use a common                    interface to store different types of data in different parts of the SAP paging             area. This data includes the ABAP/4 memory (EXPORT TO MEMORY), the                    SUBMIT REPORT parameters, CALL DIALOG and CALL TRANSACTION                        USING, as well as internally defined macros (specified with DEFINE).


We have a issue mentioned related to SAP Paging which clears more about Paging : Read Here



  1. Private memory : This is nothing but private heap memory as we give the most priority to dialog process we reserve heap memory for the dialog process. After releasing the assigned memory (difference of roll_first and roll_area ), the operating system still considers the (virtual) memory as being occupied by the allocating process. 


The work process can be run in PRIV mode (private) when the local memory is assigned, The work process is reserved for processing the current user context until the context releases the work process again when the request has ended. 


Up to this time point no other user context can be mapped to the work process.


Once the user context is terminated (work process is released, restart : restart occurs when the work process use more local memory than is defined in the parameter abap/heaplimit ) , local memory is returned. 


If a dialog work process has used up the roll area assigned to it and the extended memory, private memory is assigned to the work process. The work process goes into PRIV mode.


In the parameter settings, ensure that not too many dialog work processes are simultaneously in the PRIV mode, since this lowers system performance.


The other work process types , the non dialog work process  (background, update, enqueue and spool work processes) are assigned heap memory after the roll area is used up. Allocating Extended Memory is not useful since no context change occurs for these processes.


Since we are on this topic , lets talk about Automatic Termination of PRIV mode Processes :


If too many dialog work processes run in the PRIV mode in an SAP application server, performance problems arise. The normal processing of user inquiries is slowed down by the PRIV mode work processes.


The SAP system offers a way out that lets you terminate non-active dialog processes in the PRIV mode. 


We set rdisp/wppriv_max_no , say n . What happens after this is if you have more than this number of work process running in private mode. Work processes are active and the time span set in parameter rdisp/max_priv_time has elapsed, the transaction for that PRIV process, which has spent the longest possible time in PRIV mode, is reset



Note :-


rdisp/max_priv_time (default 600 seconds)


rdisp/wppriv_max_no. (Number of dialog work processes - 5 or 1 which ever has the grater value among this two )


Example :- 


If we set rdisp/wpriv_max_no = 5 and rdisp/max_priv_time = 600 second. Then, the SAP system checks periodically if more than 5 dialog work processes are in PRIV mode after 600 seconds (10 minutes). If this is the case, the transaction is reset that has been in PRIV mode for the longest time


Virtual Address Work Space 



 Image Source


What is Memory Allocation ?


Reserving a segment in physical memory , it makes more sense when you take the image as reference. Linking the physical memory segment to the virtual address space of the process, which means reserving a segment of the same size in the virtual address space and mapping the virtual addresses to physical addresses.


Function of SAP Memory Management


An application runs in an SAP work process where an ABAP program is normally executed. The process requires memory to do this, which is allocated to the process by the memory management system.


We have separate way of dealing with Dialog Work Process and Non Dialog Work Process.


This would be the right time for you to understand few more terminologies:-


User Context : 


Temporary memory area that contains information about users which is used throughout the program processing. User information that is stored [Current Settings of the user ,  User Authorizations , Names of program that user are currently using] , Once they log off this memory is released.


The user context area that allows direct access can be extended as required when the size of the user context increases. For dialog work processes, the data of the user context, including internal tables is located in this expanded area. You can therefore access all the data in the user context. Only data of types extract and export to memory still remain in SAP Paging


Roll Area : 


Memory area which is allocated by work processes for the program instance.

Each time a user starts a program, a roll area is created for that instance of the program.Program Information contained [ Dynamic Memory Allocations, Variable Value and Current pointer of the program ] 


The memory in the roll area becomes free when the program ends its execution.


If two users run the same program at the same time then two roll areas will be created one for each user.


The SAP Roll Area is used for the initial memory assigned to a user context, and (if available) for additional memory if the expanded memory is full.


Roll-in: cross-user data is rolled in from the common resource in the work process (and is processed there).


Roll-out: User-specific data is rolled out from the work process in the common resource (after the dialog step has ended)





Image Source (help.sap.com)


Execution of Work Process ->wp in case of Dialog and background Job



In case of Dialog work Process

In case of Background Process


  1. At first Roll_area is used : Based on Parameter Ztta/roll_first.


  1. After that we use the extended memory that is mentioned in em/intial_size_MB and use quota for particular wp which is mentioned in ztta/roll_extension_dia.


  1. After that the left roll area is used (the memory can be calculated by calculating the difference between ztta/roll_area and ztta/roll_first)

  2. Post that we use private memory [Local Heap Memory]

 

Limit of heap memory are configured using these parameters:


abap/heap_area_dia

abap/heap_area_total



  1. OS Limits allocation of Memory , swap space in the host system is used up or the upper limit of the OS address space is reached.




  1. Similarly roll area is used “ztta/roll_area” 


  1. Unlike Dialog work process extended memory are not used right now 


  1. Roll area is already used in this scenario


  1. Heap area is used , Unlike Dialog Work process Private Memory [Local Heap memory] is out of scope.


  1.  Either limit of heap memory is used based on the SAP parameter “abap/heap_area_nondia” or it uses entire heap memory assigned for all the work process and its value based on SAP parameter “abap/heap_area_total”

  2. OS limits allocation memory.

  3. SWAP space is completely used up, Rare to occur.


5. If still required is full then the extended memory is used.






Graphical Representation 


In case of Dialog process :-



In case of Non Dialog Process :-




Note :-  Once work process comes in PRIV mode it  cannot be terminated by the parameter “rdisp/max_wprun_time”, it remains connected to the user until the user ends the transactions or it is timed out by the system due to the memory or explicitly logged out using SM04/DPMON.



Read More :-


help.sap link


SAP Blog by Samar


SAP User Context and Roll Area


Working model Example

SAP Memory Management (BC-CST-MM) - SAP Library

SAP Memory Management Implementations (In Process) - Application Server Infrastructure - SCN Wiki

Memory Parameters - SAP NetWeaver Business Warehouse - SCN Wiki

Cracking SAP Memory Management _Part1

Cracking SAP Memory Management _Part2

Cracking SAP Memory Management _Part3

Memory Allocation and Analysis - Application Server Infrastructure - SCN Wiki




Comments

  1. Well done 👏 very clearly written about SAP memory management..keep it up

    ReplyDelete
  2. Please make some information on ST02 if memory swaps are increasing those things

    ReplyDelete
  3. Your way of writing and making things clear is very impressive. custom erp

    ReplyDelete

Post a Comment

You might find these interesting

Notes for Build Resilient Applications on SAP BTP with Amazon Web Services [ Week 1]

Welcome back to the next chapter in our ongoing series dedicated to unraveling the dynamic interplay between SAP Business Technology Platform (BTP) and Amazon Web Services (AWS). For those just joining us, this blog serves as an invaluable resource for individuals delving into the world of SAP BTP or seeking a comprehensive reference guide. SAP BTP, or SAP Business Technology Platform, is a comprehensive platform that brings together various essential capabilities for application development, automation, data management, analytics, planning, integration, and AI. These features are all integrated into a unified environment, making it user-friendly for both professional IT developers and citizen developers. Image Credit  Key Features of SAP BTP: Application Development: SAP BTP offers a range of tools for development. For instance, SAP Build enables low-code development, while the SAP Business Application Studio caters to core developers, providing services like document management a...

8 Must-Know Questions About Object Store on SAP Business Technology Platform

What is the problem that Object Store solves ? Modern enterprise systems increasingly deal with massive volumes of unstructured data such as documents, logs, media files, and backups. Traditional relational databases are not optimized for such workloads. What is Object Store ? Object storage—commonly referred to as blob storage—addresses this gap by providing scalable, durable, and cost-efficient storage for unstructured data. Object storage is a storage architecture designed to manage unstructured data as discrete units called objects.  Each object consists of: Binary data (file content) : Image , File etc Metadata (descriptive attributes) : File size, Content type, Last modified timestamp, Storage class (hot, cool, archive) Unique identifier (key or URL) : unique path-like string used to locate a blob inside a bucket Unlike file systems or relational databases, object storage does not rely on hierarchical file structures or schemas. The SAP BTP Object Store service is a managed, ...

Understanding SAP BTP Global Accounts, Directories, Subaccounts, and Entitlements

In SAP Business Technology Platform (BTP), organizing your resources effectively is crucial for efficient management and scalability. This blog provides a comprehensive overview of global accounts, directories, subaccounts, and entitlements within SAP BTP. What is a Global Account in SAP BTP? A global account in SAP BTP represents the contractual agreement you have with SAP. It serves as the top-level container for managing various resources, including directories, subaccounts, members, entitlements, and quotas. Within a global account, you receive entitlements and quotas for platform resources, which can be allocated to subaccounts for actual consumption. How Do Directories Function in SAP BTP? Directories in SAP BTP allow you to organize and manage your subaccounts based on your technical and business requirements. A directory can contain other directories and subaccounts, enabling you to create a hierarchical structure. This hierarchy can be up to 7 levels deep, with the global ac...

How to properly Start/Stop SAP system through command line ?

Starting/stopping an SAP system is not a critical task, but the method that most of us follow to achieve this is sometimes wrong. A common mistake that most of the SAP admins do is, making use of the 'startsap' and 'stopsap' commands for starting/stopping the system.  These commands got deprecated in 2015 because the scripts were not being maintained anymore and SAP recommends not to use them as many people have faced errors while executing those scripts. For more info and the bugs in scripts, you can check the sap note 809477.  These scripts are not available in kernel version 7.73 and later. So if these are not the correct commands, then how to start/stop the sap system?  In this post, we will see how to do it in the correct way. SAP SYSTEM VS INSTANCE In SAP, an instance is a group of resources such as memory, work processes and so on, usually in support of a single application server or database server with...

KPIs for Recovery in HANA Database Administration

Introduction: In the dynamic landscape of database administration, ensuring the robustness of a system is paramount. One crucial aspect that demands meticulous attention is the recovery process following a system failure. Two key performance indicators (KPIs) stand out in this realm – Recovery Point Objective (RPO) and Recovery Time Objective (RTO) . In this technical blog, we will delve into the significance of these KPIs for HANA database administrators and explore strategies to optimize them. Recovery Point Objective (RPO): RPO is a critical metric that defines the maximum acceptable data loss in the event of a system failure . For HANA database administrators, establishing an RPO involves a careful balance between data consistency and the overhead of continuous data replication. Continuous Data Backups: To meet stringent RPO requirements, implementing continuous data backups is imperative. Utilizing HANA's native backup capabilities and integrating them with a robust backup s...

Huge Multiversion Concurrency Control (MVCC) Versions in HANA

What is MVCC? MVCC is a database concurrency control method that allows multiple transactions to occur concurrently without conflicting with each other. In a nutshell, it ensures that each transaction sees a snapshot of the database at a specific point in time, even if other transactions are making changes concurrently. MVCC in SAP HANA: SAP HANA uses MVCC to manage concurrent access to data. Each transaction in HANA sees a consistent snapshot of the data at the time the transaction began. This is achieved by maintaining multiple versions of a data row, each associated with a specific transaction or point in time. The Issue of Huge MVCC Versions: Now, the term "Huge MVCC Versions" indicates a situation where there is a significant number of these versions for a particular set of data. Here's why this might become a problem: Increased Memory Usage: Each version of a data row consumes memory. As the number of versions increases, the overall memory consumption by the databas...

Execute HANASitter for hang situation analysis

The SAP HANAsitter is configured to perform default checks once every hour to ascertain the online and primary status of SAP HANA. Upon confirmation, it initiates tracking procedures, which involve regular responsiveness assessments (typically every minute). If SAP HANA becomes unresponsive, the HANAsitter commences recording activities, potentially capturing call stacks of active threads, run-time dumps, index server gstacks, and/or kernel profiler traces, although, by default, no recording occurs. When SAP HANA is responsive, the script scrutinizes critical features, including a standard check for more than 30 active threads. If this threshold is exceeded, the script triggers recording. Upon completing the recording process, the script exits, with an option to be configured for restart using the command line. Setup Steps Overview: Begin by creating an SAP HANA user with the desired name (e.g., HANASITTER) and assign the CATALOG READ privilege to it. Establish a user key in the hdbuse...

Deploying SAP on Google Cloud : Part 1

 Connect to Google  Connection Method To be Used for Speed Explanation Example Uses Cloud VPN Proof of Concept Variable, up to 3 Gbps Connects on-premises network to Google Cloud securely over the internet using IPsec VPN tunnels. Creating a Cloud VPN tunnel between on-premises and Google Cloud. Encrypted IPsec tunnels Dedicated Interconnect For Enterprise level connect 10 Gbps to 100 Gbps Provides a dedicated, private connection between on-premises and Google Cloud through Google's network. Provisioning a dedicated interconnect connection. Direct physical connection between on-premises and Google Cloud network infrastructure Partner Connect If you have a data center which cannot be reached to Dedicated Google facility. Variable, up to 100 Gbps Allows connecting to Google Cloud through supported service providers. Establishing a connection with a supported service provider. Utilizes service provider's network infrastructure. Configure Tunnels with Google Cloud Platform IPsec

Building the Foundation of the PO System: Architecture and some terminologies

1. Loosely Coupled and Tightly Coupled Services : Loosely Coupled Services : These services interact with each other with minimal dependencies. Changes in one service don't significantly impact others. Pros include flexibility, easier updates, and better scalability. A common example in PO is when a shipping service communicates with an inventory service. Changes in the inventory service won't necessarily disrupt shipping. Tightly Coupled Services : These services are interdependent, so changes in one service can affect others. While they might provide faster communication, they can be less flexible and harder to maintain. For example, tightly coupling an order processing service with a payment service means any change in payment could ripple to order processing. 2. SOA - Service-Oriented Architecture : SOA is an architectural approach where everything is treated as a service, encapsulating specific functionality. Service Orchestration Example (Banking Transaction) : Consider ...